• Number of bacteraemia and of fungemia - overall - each condition [ Time Frame: max Day 30 ] [ Designated as safety issue: Yes ]
  

• Number of patients with adequate anti-infective therapy [ Time Frame: at day 30 ] [ Designated as safety issue: Yes ]
  
• Adequate anti-infective therapy [ Time Frame: at 24h, 48h, > 48h ] [ Designated as safety issue: Yes ]
  
• Time between sampling for microbial investigation and positive results relevant for the diagnosis [ Time Frame: between sampling for microbial investigation and positive results ] [ Designated as safety issue: Yes ]
  
• Mortality [ Time Frame: at Day 30 ] [ Designated as safety issue: Yes ]
  
• Sepsis chock, secondary infectious focus [ Time Frame: at Day 30 ] [ Designated as safety issue: Yes ]
  
• For neutropenia cases, number of patients who evaluated with a clinical focus of infection [ Time Frame: at day 30 ] [ Designated as safety issue: Yes ]
  
• Diagnosis of endocarditis [ Time Frame: at Day 45 ] [ Designated as safety issue: Yes ]
  
• Number of non clinical investigations (microbial and non microbial) [ Time Frame: at day 30 ] [ Designated as safety issue: Yes ]
  
• Length of hospital stay [ Time Frame: at day 30 ] [ Designated as safety issue: Yes ]
  

We will evaluate the advantage of adding the molecular test to the microbial investigations usually done (blood cultures and others) in cases of febrile neutropenia, suspicion of infective endocarditis and severe sepsis in intensive care units.

This is a prospective study conducted in 18 sites (7 in the Paris area and 11 all over France) which will enrolled about 2000 patients over 18 years. Sites are randomized for starting with a 6-month period performing the test or 6-month period without the test (control time with the standard of care).

Primary outcome are the number of patients with documented bacteraemia or fungemia. Secondary outcome are (1) the number of patients with an adequate anti-infective therapy and how long it happens after the diagnosis, (2) mortality, (3) new complicated infection, (4) number of investigations (microbial and non microbial) done for the etiological diagnosis, and global hospitalization costs.

The advantage of the new test will be evaluated per protocol and with an intend to treat analyses. We hypothesized that the new test will bring 15% more microbial diagnosis than the standard of care. Consequently, and according to the number of sites interested in the study, 166 to 2500 patients will be enrolled with 480 to 750 patients with febrile neutropenia, 1000 to 1500 patients with severe sepsis in Intensive Care Units (ICU). Patients with suspicion of infective endocarditis will be evaluated for the number of diagnosis of true endocarditis according to Duke Criteria, and the time to diagnosis.

Health economic evaluation will compare the costs of hospitalization, microbial investigations including the new test, other non clinical investigations and consequences on the organization.

Inclusion Criteria:

• Age≥ 18 years
• Written signed and dated inform consent
• First time with fever observed in a neutropenic patient
• Severe sepsis in a patient hospitalized in ICU
• Suspicion of infective endocarditis
• Microbial investigation from Monday to Friday

Exclusion Criteria:

• Not affiliated to Health Insurance (social security)
• Included in another interventional trial testing microbial DNA detection during the time "without Septifast®"
• Included in another clinical trial for which the clinician assumes that it will not be possible to prescribe an anti-infectious therapy adequately to microbial detection in the blood
• Patient previously included in the protocol
• Sepsis with a microbial diagnosis